Elektrine lite

← Feed

@jam@tilde.zone

Post #1700831

2026-04-18 16:39 UTC

As it finally approaches the point where the possibility of being forced to use LLMs at work looms over me, I keep having this thought: Skill building is a secondary form of payment for labor. It doesn't put food on the table but it is a source of value you take as a part of your agreement to perform labor, either for personal benefit outside work or in future jobs. One of the reasons I largely avoid utilizing software with expensive licenses at work (even when it would make me more productive and be a rounding error in the budget) is because when I go home at the end of the day I have now lost access to some of the skill I have built with a tool, and have thus traded personal value for that of my employer. With LLMs a similar issue arises. Let's start with the assumption they are more productive (for some combination of tasks and measurements this will be true) and ignore most of the negative externalities. If I'm forced to use them, I am not building the same skills as doing the work myself, and am having a form of payment I previously was guaranteed as a part of my employment stripped from me in exchange for the benefit of my employer. Why would I agree to this? If I was told I was receiving a pay decrease of any other form why would I find that acceptable? Maybe you could argue that I'm still building skills, just different high level ones like agent orchestration or ""prompt engineering"" or whatever. But also like...that isn't what my job is? It's not in my job description? It's not something I find fun and it's not something I want to do? It's not a skill I can freely use on my own time, and it's certainly not one I want to spend my personal time doing. And if I had my way (I don't think I will) it wouldn't be a skill I'd be taking to my next job. So why am I being told I need to be paid less to do a job I don't want to do, that isn't in my employment contract, and is less enjoyable and fulfilling to me?

Replies (7)

  • @adrianco@mastodon.social 2026-04-19 08:39

    @jam@tilde.zone @cford@toot.thoughtworks.com Perhaps there are developers who like the journey and others who like the destination? Mostly people get paid to arrive at the destination using whatever tools are currently available.

    Open ##2438471

  • @tshirtman@mas.to 2026-04-19 10:54

    @jam@tilde.zone not a perfect answer to this, but there are decent open weights models that run on consumer hardware, qwen3.6-35bA3 with opencode is not claude with opus 4.7, but it's free and it is impressive, it does find non trivial bugs and fixes for them. prompt engineer is not on your job description any more than vim expert or stackoverflow reader, expectations about what software engineering do evolve with time and place. i find it really weird we test people on algorithms they'll never implement

    Open ##2438475

  • @jam@tilde.zone @linuxandyarn@hachyderm.io One piece of career advice that sticks with me is, “Don’t become good at something you hate.” This advice seems to apply to LLMs in your situation.

    Open ##2438481

  • @annaf@climatejustice.social 2026-04-19 11:16

    @jam@tilde.zone as an employer I remember reading something like ‘pay your team like they might leave, train them like they will stay’ I think that’s the correct way round, it’s meant to be a good thing! But I totally agree with you, training is part of the benefits of working somewhere.

    Open ##2438482

  • @claus@fosstodon.org 2026-04-19 11:39

    @jam@tilde.zone Maybe get another job if you don't like how your current one is changing? Happens in many jobs and different industries..

    Open ##2438483

  • @poleguy@mastodon.social 2026-04-19 12:32

    @jam@tilde.zone this is a very interesting post! I largely agree. I hate using LLMs for various reasons. But I actually find that using an LLM to write, analyze and debug code is increasing my exposure to techniques that I wouldn't have been exposed to b otherwise, and provides training in a way that had been lacking in the past. I hear llm described as an intelligence amplifier. I think it also amplifies learning too, provided you actually spend the time to study the methods and techniques it surfaces.

    Open ##2438485

  • @arildsen@mastodon.green 2026-04-19 14:55

    @jam@tilde.zone I have worried for some time now that using LLMs extensively would lead to my own skills atrophying, but I hadn’t really thought about the training as job benefit aspect. That’s a good point. I am a still a bit on the fence about my skills, because using LLMs for various programming tasks also teaches me about additional techniques, libraries etc.

    Open ##2438488